🚀 go-pugleaf

In article <1991Mar14.191513.20184@iitmax.iit.edu> gkt@iitmax.iit.edu (George Thiruvathukal) writes:
>In article <STUCKI.91Mar14120723@amygdala.cis.ohio-state.edu>, stucki@amygdala.cis.ohio-state.edu (David J Stucki) writes:
>> 	The  use of non-standard libraries should  not  be a disadvantage,
>> since the  "standard" ones are  rarely implemented correct  wrt Wirth.
>> Cooper's IO module is much cleaner than either the InOut provided by 
>> Metrowerks (for the MAC) or Sun.
>
>I agree with the principle stated above; however, non-standard libraries ought
>to have a standard implementation.  Without a standard set of libraries or,
>minimally, a library which is implemented in the standard language, how can
>any degree of portability between implementations be achieved?  It cannot be
>achieved.
>
>It is somewhat regrettable that Niklaus Wirth left the issue of libraries so
>open-ended that programmers cannot rely upon a minimal set of libraries from
>implementation to implementation.  Even in the C language, one can count on
>input/output functions.  As the above excerpt suggests, one should not be able
>to count on input/output (something so fundamental) being well-implemented by
>compiler vendors.  Would it not make some sense that one ought to question the
>abilities of a compiler writer who cannot design and implement simple libraries
>to accompany his/her compiler?  Further, if the standard libaries cannot be 
>implemented correctly, how on earth is a variation on the same theme to be
>implemented?
>-- 
>George Thiruvathukal
>
>Laboratory for Parallel Computing and Languages
>Illinois Institute of Technology
>Chicago
Is it any wonder that Fortran and Cobol don't go away?  Back in the "dark ages"
I/O was an integral, standardized, part of the language.  In Ada, I/O (and
tasking) are disguised to look like procedures, but at least they're standard-
ized.


-- 
vsnyder@jato.Jpl.Nasa.Gov
ames!elroy!jato!vsnyder
vsnyder@jato.uucp